Nematinae Thomson 1871

Monophyly and delimitation of Nematinae Based on ten genes (COI and 9 nuclear genes), the subfamily Nematinae is strongly supported as monophyletic (Fig. 1), as found in previous molecular phylogenetic analyses with adequate taxon and gene sampling (Prous et al. 2014; Malm & Nyman 2015; Vilhelmsen et al. 2024; Wutke et al. 2024). Support for monophyly of Nematinae based on morphology is ambiguous(Vilhelmsen 2015). Unusually, some genera of Nematinae (Analcellicampa, Caulocampus, Cladius, Hoplocampa, Monocellicampa, and Susana) are classified under a separate subfamily Hoplocampinae Rohwer, 1911 in NCBI GenBank. This arrangement of genera (Hoplocampinae) finds no support in any phylogenetic analyses (Prous et al. 2014; Malm & Nyman 2015; Vilhelmsen 2015; Fig. 1), except perhaps a classification proposal by Wei & Nie (1998), which is not widely accepted (Taeger et al. 2010). The genera of Hoplocampinae are strongly supported as part of Nematinae radiation, but the exact relationships between many of these and the other genera of Nematinae are poorly supported (Prous et al. 2014; Malm & Nyman 2015; Niu et al. 2021; Fig. 1). The best supported clades within the Nematinae are Dineurini Ashmead, 1898 (Prous et al. 2014) and Nematini (Figs 1–2), the latter containing the majority of species of Nematinae (more than 75%), including the genus Euura. Phylogeny and groups of Euura The core of the genus Euura (i.e., without the Euura epilosa group) is strongly supported as monophyletic based on phylogenetic analyses of ten genes (Fig. 2). The Euura epilosa group (sampled species E. brachycera and E. epilosa) is weakly or moderately supported as the sister group of the core of Euura (Fig. 2). The Nematus wahlbergi group is found as the sister group of Euura with moderate support, rendering Nematus paraphyletic (Fig. 2). Nematus together with E. brachycera (DEI-GISHym86097 as the only representative of the group which was identified as Nematus spec.) formed a weakly supported sister clade of core Euura in a previous analysis based on a smaller dataset (Prous et al. 2019). Euura brachycera (Lindqvist, 1975) (Euura brevicera Taeger & Blank, 2014 is an unnecessary replacement name) was transferred to Euura by Prous et al. (2014), although no genetic data was available for this or other related species at the time. Even though the Euura epilosa group is not genetically strongly supported as a member of Euura, we retain it in this genus because of its similar morphology [see under Euura epilosa (Lindqvist, 1970)]. To reconstruct internal phylogeny of Euura, we excluded mitochondrial COI, because of strong conflict with nuclear genes: in some cases even above group level (see under Mitonuclear discordance). For most species we included 1– 2 specimens having the greatest amount of sequence data, but more specimens in some cases because of large genetic variability (Fig. 3). At the base of core Euura (excluding the Euura epilosa group) 15 clades are well-supported, but relationships among them are poorly resolved (Fig. 3). In combination with morphology, we have defined 37 species groups, 7 of which are gall-makers (Liston et al. 2017), and 12 species not assigned to groups. The largest well-supported clade (clade A) includes roughly 30% of species of Euura, containing most of the former “ Amauronematus ” (including former “ Pontopristia ”, but excluding “ Brachycoluma ”), i.e., the punicea group sensu lato (s. lat.), and flavescens, scutellata, miliaris, melanocephalus, vicina, caeruleocarpus groups, and some species not assigned to groups (imperfecta, salicis, distinguenda, anthracina, hyperborea) (Fig. 3). Within this clade, the punicea group s. lat. is strongly supported, containing the former “ Amauronematus ” groups: aegra, amentorum, histrio, humeralis, longiserra (not monophyletic based on nuclear data), opacipleuris, punicea sensu stricto (s. str.), toeniata, tunicata, variator, vittata, and E. brunnea not assigned to a group. Clade A also includes a well-supported clade comprising the miliaris, scutellata, and flavescens groups, where the last two may be sisters. The second largest well-supported clade comprises the gallmakers, followed by three other well or moderately supported clades containing two groups each: miltonota + viduata, myosotidis + bipartita, and pallescens + ribesii (Fig. 3). A neighbour-joining distance tree including specimens with minimum of 1529 bp of nuclear data (1381 specimens) is included as supplementary material. Within-species genetic diversity in Euura Without any knowledge of species boundaries, it is possible to get information about within-species genetic diversity based on haplotype divergence of nuclear genes of single individuals. As Hymenoptera are haplodiploid (males are haploid, females diploid), this assessment is normally only possible for female individuals (although rare diploid males exist). Averaging across individual females with at least 1529 bp of nuclear data, the average divergence between the haplotypes in Euura is 0.32% (max 1.41%). When considering all individuals of each species as delimited here, the average within-species divergence is 0.23%. This is counter-intuitive, as within-individual haplotype divergence should underestimate within-species divergence; however, this result is because we coded heterozygous positions using IUPAC ambiguity symbols, which are treated as missing data in the calculations. When only males are considered (i.e., hemizygous, therefore no ambiguous positions), average within-species divergence is 0.42%, which is larger than estimated from within-individual female haplotypes. Consistent with morphological observations, the between-species genetic divergence within species groups is often only slightly larger (on average varies between 0.5–2.1%) than within-species divergence, highlighting the difficulty of delimiting species in Euura. Mitonuclear discordance As in Pristiphora (Prous et al. 2017), in Euura there is a strong conflict between mitochondrial COI on one hand, and nuclear and morphological data on the other hand when delimiting species – or even groups above species level in some cases (see supplementary distance COI and nuclear trees). Currently, over 50% of the species in Euura cannot be reliably identified using COI sequences. However, the precision of identification could probably be increased by comparing exact COI variants (i.e., differing by at least one nucleotide), which often correlate with species boundaries. At present, due to the large diversity of COI variants, the correspondence between exact variants and species has not yet been established in most cases. Many additional COI variants (hundreds, if not thousands) can be expected in Finland alone, where more than 4500 Euura specimens have been barcoded. Even ignoring this challenge, different species frequently have identical sequences: based on current data (>1300 specimens for which nuclear genes are also available) this is the case for 47 species using just the standard barcode region, or 42 species using the full 1078 bp. This is more than 20% of the species, not counting gall-makers. Identification success using COI in Euura may be higher in southern and central Europe because of lower species richness and lower within-species genetic diversity. Notably, in about 15% of species, genetic diversity is sufficiently high that species identification based on 1–2 nuclear genes can also be ambiguous. The success rate increases when more nuclear genes are included. Remarkably, in some cases there is strong disagreement between mitochondrial COI and nuclear genes for delineating groups of species. The most remarkable is the non-monophyly of the E. amentorum group based on COI, contrasting with strong support based on nuclear genes (Liston et al. 2023). Another example, albeit less extreme, is the clade comprising the E. histrio, E. variator, and E. opacipleuris groups, wherein groups can be delimited well by morphology and nuclear genes, but COI sequences of the same species can in several cases appear to belong to different groups. There are a few other such examples throughout Euura (see supplementary distance COI and nuclear trees of the specimens analysed here). Intraindividual mitochondrial variants in Nematinae As already reported (Prous et al. 2021; Liston et al. 2023), multiple intraindividual COI variants have been detected in many Nematinae using Nanopore sequencing. The vast majority of these variants appear to be functional, because they lack stop codons or frame shifting indels (insertions or deletions). This would suggest heteroplasmy, rather than the presence of nuclear-encoded mitochondrial pseudogenes (NUMTs). Excluding the detectable NUMTs, about 25% of about 1000 specimens still have more than one apparently functional variant (1078–1087 bp amplicons). It is possible that some of the detected intraindividual variants are PCR chimeras (if there is mixture of different sequences, DNA polymerase sometimes switches the template molecule during PCR), but this seems to be a minor issue. This is indicated by the fact that for nuclear genes of males (which are haploid) almost always only a single variant is detected (contrasting with frequent multiple COI variant detection) and for nuclear genes of females (diploid) usually two variants are detected (sometimes one, almost never more than two), again contrasting with frequent detection of more than two variants for COI. We did notice possible PCR chimeras among single Nanopore reads and sometimes detected such consensus variant sequences, but these have usually been of sufficiently low frequency that the variants are not detected or the variants have been filtered out because of low read counts. The frequent presence of multiple COI variants within the same individual has also been evident in the results of Sanger sequencing (Prous et al. 2021), although this method does not allow determination of the number and